Bayclone: Bayesian Nonparametric Inference of Tumor Subclones Using NGS Data

Abstract

In this paper, we present a novel feature allocation model to describe tumor heterogeneity (TH) using next-generation sequencing (NGS) data. Taking a Bayesian approach, we extend the Indian buffet process (IBP) to define a class of nonparametric models, the categorical IBP (cIBP). A cIBP takes categorical values to denote homozygous or heterozygous genotypes at each SNV. We define a subclone as a vector of these categorical values, each corresponding to an SNV. Instead of partitioning somatic mutations into non-overlapping clusters with similar cellular prevalences, we took a different approach using feature allocation. Importantly, we do not assume somatic mutations with similar cellular prevalence must be from the same subclone and allow overlapping mutations shared across subclones. We argue that this is closer to the underlying theory of phylogenetic clonal expansion, as somatic mutations occurred in parent subclones should be shared across the parent and child subclones. Bayesian inference yields posterior probabilities of the number, genotypes, and proportions of subclones in a tumor sample, thereby providing point estimates as well as variabilities of the estimates for each subclone. We report results on both simulated and real data. BayClone is available at

Nonparametric Bayesian Inference in Biostatistics

“…Chapter 4 (Ji et al 2015) uses an Indian buffet process as a prior probability model for a feature allocation problem. Some of the chapters use models beyond this selection.…”

“…The limit of the model, as C → ∞ becomes a constructive definition of the IBP (Griffiths and Ghahramani 2005;Teh et al 2007). To account for a biologically more accurate description, taking these details into account, Lee et al (2015b) proposed a linked feature allocation model based on cIBP (Sengupta 2013;Sengupta et al 2015), which we review next. Note that m c = 0 is possible with positive prior probability.…”

Bayesian Inference for Intratumour Heterogeneity in Mutations and Copy Number Variation

“…Assuming a priori independence among subclones, we write . Similarly to Griffiths and Ghahramani's () constructive definition of the IBP, a limit of the described model p ( L ) becomes a definition of a CIBP under the following construction (Sengupta, ; Sengupta et al ., ). Let β =1 and C →∞ and drop all columns with all 2s.…”

“…() and Sengupta et al . () in that it accounts for the mean number of copies of locus s in sample t . We consider with to inform a small value a priori .…”

“…() and Sengupta et al . () which implements inference for TH based on SNV data only, and therefore using a model with Z and w only, implicitly assuming that all loci are copy number neutral. An important implication of the different modelling approaches is that inference in Lee et al .…”